A hair cell-specific peroxidase coordinates stomatal and prickle hair cell size to optimize epidermal cell functionality in grasses

The leaf epidermis is the outermost cell layer forming the interface between plants and the atmosphere that must accomplish contrasting functions; it must provide a robust barrier against biotic and abiotic stressors while simultaneously facilitating efficient exchange of carbon dioxide and water. To achieve these opposing requirements, the plant epidermis developed a wide range of specialized cell types such as stomata and hair cells (=trichomes). While factors forming these individual cell types are known, it is poorly understood how their number and size is coordinated. Here, we identified a role for BdPRX76/BdPOX, a class III peroxidase, in regulating both stomata and hair cell size in the model grass Brachypodium distachyon. In bdpox mutants stomata were longer, yet prickle hair cells were smaller. Because stomatal density remained unchanged, the negative correlation between stomatal size and density was disrupted and resulted in higher stomatal conductance and lower intrinsic water-use efficiency. Reporter lines revealed that BdPOX was exclusively expressed in hair cells suggesting that BdPOX cell-autonomously promotes hair cell size and non-cell-autonomously restricts stomatal length. Cell wall autofluorescence and lignin stainings indicated a role for BdPOX in lignification or crosslinking of related phenolic compounds at the hair cell base. Ectopic expression of BdPOX in the stomatal lineage increased phenolic content in guard cell walls and restricted stomatal elongation. Together, we reveal a developmental coordination between hair cells and stomata that optimizes epidermal functionality. We propose that a cell-type-specific disruption of this coordination leads to compensatory developmentaldefects in other epidermal cell types.


RESULTS
The spatial organization and functionality of leaf epidermal cells play fundamental roles in plant productivity and (a)biotic interactions (Dow et al., 2014;Javelle et al., 2011;Riglet et al., 2021;Zuch et al., 2022). In the model grass B. distachyon, the leaf blade epidermis is dominated by rectangular pavement cells, stomatal complexes, trichomes consisting of prickle hair cells (PHCs) and few interspersed macrohairs, and occasional silica cells (Fig. 1A) (Hall et al., 2020). Most specialized amongst epidermal cells are stomatal complexes, which are cellular pores on the epidermis that regulate carbon capture for photosynthesis and limit water loss. Grasses present extremely fast stomatal movements, which seem to be linked to their innovative, graminoid morphology consisting of two dumbbell-shaped guard cells (GCs) flanked by two subsidiary cells (SCs) (Franks and Farquhar, 2007;McAusland et al., 2016;Nunes et al., 2020). Fast stomatal movements and adjustments in stomatal anatomy (stomatal density and size) are crucial for water-use efficient gas exchange thereby contributing to abiotic stress resilience (Lawson and Vialet-Chabrand, 2019;Nunes et al., 2022). Still, how form and function of grass stomata are regulated remain elusive.

bdpox mutants show increased stomatal conductance and cell size defects in stomata and prickle hair cells
To identify novel factors associated with graminoid stomatal morphology and functionality, we performed RNA-sequencing of mature zones of 7 day old leaves in B. distachyon Bd21-3 (WT) and bdmute plants (Fig. S1A). The bdmute leaf epidermis features abnormal stomata that lack SCs which strongly affects stomatal responsiveness and gas exchange (Raissig et al., 2017). 179 genes were downregulated in bdmute ( Table 1 in Supplementary dataset S1) and we selected ~50 candidate genes for a reverse genetic screening. Candidates were chosen according to their annotated gene function, their A hair cell-specific peroxidase coordinates stomatal and prickle hair cell size to optimize epidermal cell functionality in grasses The leaf epidermis is the outermost cell layer forming the interface between plants and the atmosphere that must accomplish contrasting functions; it must provide a robust barrier against biotic and abiotic stressors while simultaneously facilitating efficient exchange of carbon dioxide and water. To achieve these opposing requirements, the plant epidermis developed a wide range of specialized cell types such as stomata and hair cells (=trichomes). While factors forming these individual cell types are known, it is poorly understood how their number and size is coordinated. Here, we identified a role for BdPRX76/BdPOX, a class III peroxidase, in regulating both stomata and hair cell size in the model grass Brachypodium distachyon. In bdpox mutants stomata were longer, yet prickle hair cells were smaller. Because stomatal density remained unchanged, the negative correlation between stomatal size and density was disrupted and resulted in higher stomatal conductance and lower intrinsic water-use efficiency. Reporter lines revealed that BdPOX was exclusively expressed in hair cells suggesting that BdPOX cell-autonomously promotes hair cell size and non-cell-autonomously restricts stomatal length. Cell wall autofluorescence and lignin stainings indicated a role for BdPOX in lignification or crosslinking of related phenolic compounds at the hair cell base. Ectopic expression of BdPOX in the stomatal lineage increased phenolic content in guard cell walls and restricted stomatal elongation. Together, we reveal a developmental coordination between hair cells and stomata that optimizes epidermal functionality. We propose that a cell-type-specific disruption of this coordination leads to compensatory developmentaldefects in other epidermal cell types.  .2G0467800) gene model indicating the location and nature of the mutations in bdpox-1 (NaN1508) and bdpox-2 (NaN1528). (C) Stomatal conductance (g sw ) in response to changing light (1000-100-1000-0 PAR) in WT, NaN1508 BdPOX , bdpox-1 and bdpox-2 (n=6 individuals per genotype). Dots represent the mean and error bars represent SEM. (D) Intrinsic water-use efficiency (iWUE) in response to changing light (1000-100-1000 PAR) in WT, NaN1508 BdPOX , bdpox-1 and bdpox-2 (n=6 individuals per genotype). Dots represent the mean and error bars represent SEM. (E) Stomatal density (SD) in WT, NaN1508 BdPOX , bdpox-1 and bdpox-2 (n=6 individuals per genotype, 548-619 stomata counted per genotype). (F) Stomatal length (SL) in WT, NaN1508 BdPOX , bdpox-1 and bdpox-2 (n=6 individuals per genotype; 193-287 stomata per genotype). Inset shows a DIC image of WT and bdpox-1 stomata; GCs are false-colored in green; scale bar, 10 µm. (G) Correlation between average stomatal length (SL) and average stomatal density (SD) in WT and bdpox-1 (n=33-. expression being lower in the developmental zone (Zhang et al., 2022) and the availability of mutants from a collection of sodium azide (NaN 3 ) mutagenized and fully resequenced lines (Granier et al., 2016).
To confirm the gas exchange defects in bdpox mutants, we measured stomatal conductance (g sw ) in response to changing light intensity (1000-100-1000-0 PAR) . In bdpox plants, we observed higher g sw in all light steps (Fig. 1C), but no significant impact on stomatal opening and closure speed ( Fig. S2A-C). Since no significant variation in carbon assimilation (A) was observed (Fig. S2D), bdpox mutants suffered a decrease in intrinsic water-use efficiency (iWUE, A/ g sw ), particularly in the light-limited step (100 PAR; Fig. 1D).
To test if the increased g sw was caused by changes in stomatal density we performed microscopic analysis of the leaf epidermis from the leaves assessed for gas exchange. No differences were found regarding stomatal density (SD; Fig. 1E), yet we observed significantly longer stomata in bdpox mutants (Fig. 1F). This suggested that the well-established negative correlation between stomatal size and density, observed both interspecifically (de Boer et al., 2016;Doheny-Adams et al., 2012;Zhang et al., 2021) and intraspecifically (Dittberner et al., 2018;Nunes et al., 2022;Sun et al., 2014), was disrupted in bdpox (Fig. 1G). Detailed morphometric confocal analysis of fusicoccin-treated leaves to induce full stomatal opening revealed that stomatal pores are indeed longer and larger in bdpox ( Fig. S2F-I).
To demonstrate if the higher g sw levels could be explained by the disrupted stomatal anatomy in bdpox, we compared physiological g s max measurements with anatomical g s max calculations (theoretical g s max based on gas diffusion physical constants and stomatal anatomical traits) using an established equation recently optimized for grass stomata . Physiological g s max measurements confirmed the increased g sw capacity in bdpox mutants (Fig. 1H), and anatomical g s max calculations revealed the same relative variation between bdpox mutants and WT (Fig. 1I). Together, this strongly suggested a causal relationship between longer stomata and higher gas exchange in bdpox.
To verify if the cell size defect was specific to stomata, we also measured the length of pavement cells (PCs) and of the base of prickle hair cells (PHCs). While no differences were found in PC size among genotypes (Fig. S2J), we observed an unexpected decrease in the base length of PHCs in bdpox mutants (Fig. 1J).
In summary, BdPOX seemed to negatively regulate stomatal size but positively regulate PHC size.
Ubiquitous expression of BdPOX in bdpox1, however, had no effect on cell size nor on gas exchange suggesting that a cell type-specific expression is necessary for its complementation ( Fig. S3F-I).
Together, our results suggested that BdPOX played a cell-autonomous role in promoting PHC size and, as a consequence, non-cell autonomously restricted stomatal elongation.

BdPOX might be involved in lignification/hydroxycinnamates cross-linking at the base of prickle hair cells to positively regulate cell outgrowth
To mechanistically link how a hair-cell localized peroxidase could affect stomatal anatomy, we first investigated the cell-autonomous function of BdPOX in PHCs. The Arabidopsis homolog of BdPOX, AtPRX66, is associated with phenolic modifications in the cell wall, namely lignification of tracheary elements (Tokunaga et al., 2009). A unique feature of grass cell walls is the significant yet cell-type-specific amount of hydroxycinnamates (ferulic acid and ρ-coumaric acids) that are bound to arabinoxylans and/or to lignins (Coomey et al., 2020;Hatfield et al., 2016;Rancour et al., 2012;Vogel, 2008). The polymerization and crosslinking of these compounds (including monolignols into lignins) are catalyzed by laccases and/or class III peroxidases (Burr and Fry, 2009;Hoffmann et al., 2020;Le Bris et al., 2019;Rojas-Murcia et al., 2020;Wang et al., 2015). Thus, we hypothesized that Bd-POX modulates PHC size by altering phenolic compounds in the cell walls.
To test this, we assessed UV-induced autofluorescence of phenolic compounds in the cell wall of PHCs (Donaldson, 2020;García-Plazaola et al., 2015;Shtein et al., 2017). PHC autofluorescence plot profiles and corrected total cell fluorescence revealed lower autofluorescence in bdpox-1 compared to WT and complemented bdpox-1, specifically at the base of the PHCs (approximately the initial 12 µm; Fig. 3A To test if the lower cell wall autofluorescence originated from decreased lignin/hydroxycinnamates content, we used different histochemical stainings. Basic fuchsin is a standard lignin stain (Holzwart et al., 2018;Sexauer et al., 2021;Ursache et al., 2018) that also has a high affinity for hydroxycinnamates in the B. distachyon cell wall (Kapp et al., 2015). Simultaneous imaging of cell wall autofluorescence and fuchsin-stained lignin showed that fuchsin preferentially stained the lower section of PHCs while total phenolics autofluorescence was observed until the tip ( Fig. 3D-F; Movie S3, Movie S4). Indeed, bdpox-1 mutants showed lower fuchsin fluorescence intensity at the basal section of PHCs (first 12 µm from the basal outline) compared to WT and complemented bdpox-1 lines (Fig. 3F, G; Movie S3-S5) suggesting re-duced lignin/hydroxycinnamates content in the mutant. Very similar results were observed using safranin-O lignin staining ( Fig. S4B, C), in which increased red fluorescence is observed in lignified cells, whereas non-lignified cell walls preferably fluoresce in green (Bond et al., 2008). Therefore, a red/green ratiometric analysis allows for a semi-quantitative evaluation of cell wall lignification (Baldacci-Cresp et al., 2020). bdpox-1 mutants displayed a lower safranin-o ratio at the base section of PHCs (first 12 µm) compared to WT and complemented bdpox-1, again indicating a decrease in lignin content in the mutant (Fig. S4B, C).
Regarding stomata, no significant differences were observed in autofluorescence nor in fuchsin-stained lignin in mature GCs between WT and bdpox-1 (Fig. S4D, E). When looking at the developing stomata during stomatal elongation/maturation, we observed that cell wall autofluorescence increased in wild-type GCs (Fig. S4F). This increase appeared to start earlier in bdpox-1 but stalled sooner, too, which may be linked to the stomatal elongation defects in the mutants (Fig. S4F).
Overall, our data suggests that BdPOX participates in lignification of the basal cell wall of PHCs, which seems to be required for proper PHC growth, and indirectly impacts stomatal elongation.

Specific expression of BdPOX in the stomatal lineage arrests stomatal elongation
BdPOX appeared to be involved in cell wall phenolic modifications (lignification/crosslinking) at the base of PHCs to promote cell growth. How this PHC-specific process, however, affected stomatal elongation remained elusive. When following the basipetal developmental gradient of the B. distachyon epidermis, we found that PHCs grow and mature significantly before the stomatal complexes start to elongate (Fig. 4A). PHC outgrowth started when stomata were in early stages of development (i.e. stage 3-4 during SC recruitment) and was completed before GCs elongated and acquired the mature dumbbell morphology (Fig. 4A). Therefore, the growth restriction of PHCs in bdpox could secondarily influence stomatal anatomy, but not vice versa. Accordingly, the ectopic expression of BdPOX in the developing GC lineage using a stomatal-lineage specific promoter (BdMUTEp:Bd-POX(CDS)-mCitrine; Fig. 4B) significantly restricted stomatal elongation in bdpox-1, which correlated with increased phenolics autofluorescence in GCs (Fig. 4C, D). PHC size, however, remained unaffected (Fig. 4E) as PHCs matured before stomata elongated. Intriguingly, we observed aberrant cell divisions in the pavement cells surrounding stomata when expressing BdMUTEp:BdPOX(CDS)-mCitrine in bdpox-1 (Fig. 4F, G). This suggested that elongating stomata might compensate for tissue-wide mechanical imbalances caused by too small PHCs in the bdpox-1 epidermis. Thus, when ectopic, stomatal lineage-specific expression of BdPOX inhibited compensato-ry stomatal elongation in the bdpox-1, additional divisions in pavement cells around stomata might be triggered to compensate for these tissue-wide mechanical tensions instead.
In conclusion, ectopically expressed BdPOX in GCs can modify phenolic cell wall content and modulate cell size. We propose that the size of stomata and PHCs must be coordinated for optimal epidermal functionality and that disturbing this coordination leads to compensatory developmental defects.

DISCUSSION
The leaf epidermis is the barrier between the inner photosynthetically active tissues and the environment. Highly specialized epidermal cell types facilitate the contrasting functional requirements of this outermost barrier like preventing desiccation, enabling gas exchange or deterring herbivores. Stomata, for example, open to enable carbon dioxide uptake and close to limit water loss. Trichomes, on the other hand, protect the leaf from (a)biotic stressors (Javelle et al., 2011). Several studies suggest that trichome patterning intersects with the core stomatal developmental programs in Arabidopsis (Adrian et al., 2015;Kanaoka et al., 2008;Torii, 2021;Yan et al., 2014). Also in grasses, failing to specify stomatal identity results in hair cells being formed in their place in B. distachyon (Raissig et al., 2016) and failing to specify hair cell identity leads to ectopic stomata in maize (Kong et al., 2021). This suggests that these cell types are ontogenetically closely related in grasses and that their development is thus likely coordinated. Furthermore, stomatal and trichome densities were shown to be negatively correlated in Solanum lycopersicum. The stomata to trichome ratio determined water-use efficiency (Galdon-Armero et al., 2018) suggesting a physiological relevance for the coordination between the two cellular features. Yet, the mechanisms that coordinate the formation and growth of stomata and trichomes remain highly unexplored.
Many core players guiding grass stomatal de- velopment were characterized (Abrash et al., 2018;Cartwright et al., 2009;Facette et al., 2015;Humphries et al., 2011;Jangra et al., 2021;Liu et al., 2009;Nunes et al., 2020;Raissig et al., 2017Raissig et al., , 2016Wang et al., 2019;Wu et al., 2019), whereas hair cell formation remains poorly explored (Torii, 2021). Few grass trichome initiation factors such as the transcription factors HAIRY LEAF 6 (HL6) (Fei et al., 2020;Sun et al., 2017)(Angeles-Shim et al., 2012Sun et al., 2017) and SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 10/14/26 (SPL10/14/26) (Kong et al., 2021) , (Lan et al., 2019; were identified in rice and maize, but factors that affect morphogenesis and size remain mostly unknown. Here, we identified BdPOX and revealed its role in coordinating PHC and stomatal size for optimal water-use efficiency in the mod-el grass B. distachyon. BdPOX was exclusively expressed in the hair cells and seemed to participate in the lignification or/ and cross-linking of cell wall phenolics (such as hydroxycinnamates) specifically at the base of PHCs. Since lignin is a cell wall polymer that provides mechanical support (Li et al., 2009;Rogers and Campbell, 2004;Zhao et al., 2020) we suggest that such cell wall modifications at the base of PHCs are required to increase tensile strength and provide physical support for cell outgrowth.
The reduction in PHC size in bdpox non-cell-autonomously altered stomatal size, but did not translate to changes in stomatal density. The resulting disruption of the negative correlation between stomatal size and density likely has two reasons; first, changes to stomatal size occur much later in development than the determination of stomatal density. Thus, changes in stomatal density can affect stomatal size a posteriori, where an increase in stomatal numbers can induce a downstream effect on the cell-wall machinery controlling stomatal elongation. This process is very unlikely to happen in the other direction particularly in grasses, where early stages are not only temporally but also spatially separated from late stages. Second, bdpox primarily impacts the PHC lineage and only affects stomatal development as a secondary consequence. Without a disruption of the stomatal genetic toolbox itself it is unlikely that a compensatory mechanism could be induced in a timely manner.
However, the exact mechanism of how restricting PHC growth induces stomatal elongation remains vague. We speculate that decreased PHC size may lead to changes in mechanical and/or geometrical constraints in the epidermal tissue, which would allow for increased stomatal elongation as a compensatory mechanism to reconstitute the tensile balance in the epidermis. The increase in stomatal length observed in bdpox mutants (~9 %) was quantitatively equivalent to the decrease in PHC base length (~10 %). In addition, expressing BdPOX in the GC lineage of bdpox-1 resulted in an epidermis containing both shorter stomata and shorter PHCs and induced aberrant cell divisions surrounding the stomatal complexes. We speculate that the combination of restricted PHC growth due to bdpox-1 and a prevention of stomatal elongation due to GC-expressed BdPOX may have caused a mechanical imbalance in the elongating epidermis resulting in cell divisions to compensate cellular tensions particularly around stomatal complexes. Alternatively, changes in hydrogen peroxide levels in the PHC apoplast due to loss of BdPOX might affect the reactive oxygen species signaling landscape, which could influence stomatal length non-cell-autonomously.
Regardless, the unique disruption in the negative correlation between stomatal size and density allowed us to investigate how modifying a single stomatal anatomical trait (i.e. stomatal size) would affect gas exchange. While an increase in stomatal size enhanced stomatal conductance, it did not significantly affect stomatal opening and closing speed, corroborating our previous observation that stomatal speed was correlated with stomatal density but not with stomatal size in B. distachyon .
Overall, we identified a hair cell marker that coordinates stomata and PHC formation for optimal leaf epidermal functionality. In addition, BdPOX provides a novel tool to explore the molecular identity of PHCs and help to further elucidate the developmental and functional interplay between PHCs and stomata. We highlight a way to change length, cell-wall composition, and potentially even stiffness of grass trichomes that are of major relevance for biotic and abiotic interactions, namely in protection against herbivory and high sun irradiance (Andama et al., 2020; González Moreno et al., 2022;Hall et al., 2020;Karabourniotis et al., 2020). Finally, we revealed an unexpected, indirect route to engineer stomatal size in grasses without affecting stomatal density by changing cell wall properties and growth of PHCs.

DATA AND CODE AVAILABILITY
All gas exchange and microscopic quantifications used in this study can be found in Supplementary Dataset S2. Raw and processed RNA-sequencing data are available at Gene Expression Omnibus (GEO) with the accession number GSE206682. Images are available upon request.

ACKNOWLEDGMENTS
The authors thank Prof. Dr. Annika Guse and Prof. Dr. Jochen Wittbrodt for access to microscope facilities, Prof. Dr. Karin Schumacher, Prof. Dr. Rüdiger Hell, Dr. Upendo Lupanda, and Dr. Paula Ragel for the scientific and technical insights and encouragement on the development of the project. We furthermore thank Michael Schilbach for managing the greenhouse and for gardening support. We would also like to acknowledge Prof. Dr. Dominique C. Bergmann and the Howard Hughes Medical Institute for supporting the RNA-sequencing experiment and Laura R. Lee for help with RNA-seq data analysis.

DECLARATION OF INTERESTS
The authors declare no potential conflicts of interest.

Supplementary Dataset 2:
Leaf-level gas exchange data, anatomical data from light and confocal microscopy and data from histochemical experiments (autofluorescence and lignin stainings).

Transcriptional profiling of WT and bdmute leaves by RNA-sequencing
25 mature leaf zones (25-30 mm from the base of 2 nd leaves) per replicate were collected from three wild-type (Bd21-3) replicates and from three bdmute replicates (7 days after germination seedlings grown on ½ MS plates at 20 o C with ~100 µmol photons m -2 s -1 light) were carefully collected, snap-frozen in liquid nitrogen and grounded using mortar and pistil. RNA extraction, library preparation, RNA-sequencing and data analysis was essentially performed as described in (Zhang et al., 2022). To be specific, total RNA was isolated using Qiagen's RNeasy Plant Mini kit with on-column DNAse digestion according to the manufacturer's instructions. The Kapa mRNA HyperPrep (Roche) was used to generate an mRNA enriched sequencing library with an input of 1µg of total RNA. The libraries were sequenced using the Illumina NextSeq500 platform. Read quality was assessed with FastQC and mapped against the Bd21-3v1.0 genome using bowtie2. Mapped reads were counted using summarized overlap and differentially expressed genes were analyzed using DeSeq2. Finally, gene expression was normalized by transcripts per kilobase million (TPM). Raw and processed data are available at Gene Expression Omnibus (GEO) with the accession number GSE206682.
Growth conditions for B. distachyon are specified in . In short, plants on soil were grown in a greenhouse with 18h light:6h dark cycle (200-400 μmol m −2 s −1 ; day temperature = 28 o C, night temperature = 22 o C).
To clone the BdPOX genomic sequence, a point mutation was induced in the genomic BdPOX sequence to eliminate a BsaI/Eco31I site (GGTCTC) in the second intron. Genomic BdPOX was amplified in two separate fragments using priTN99/priTN102 and priTN100/priTN101 (containing the bp substitution AGTCTC). The two resulting PCR products were purified using the NucleoSpin Gel and PCR Cleanup kit (Ref. REF 740609.50;Düren,Germany), digested at 37 o C overnight using FastDigest Eco31I (Thermo Fisher Scientific, Waltham, Massachusetts, USA) and ligated overnight at 16 o C using T4 ligase (NEB, Ipswich, Massachusetts, USA). The fully reassembled BdPOX gene was ligated overnight at 16 o C with the previously digested (Fast-Digest Eco31I) and dephosphorylated pGGC000 entry vector (with Antarctic Phosphatase, NEB) to generate pGGC_Bd-POX. BdPOX (CDS) sequence was amplified using priTN99/ priTN100 from cDNA. The resulting PCR product was purified and digested using FastDigest Eco31I (Thermo Fisher Scientific), and ligated using T4 ligase (NEB) with previously digested (FastDigest Eco31I) and dephosphorylated (Antarctic Phosphatase) pGGC000 entry vector to generate pGGC_ BdPOX(CDS).

Generation and Analysis of Transgenic Lines
Embryonic calli derived from Bd21-3 and bdpox-1 parental plants were transformed with AGL1 Agrobacterium tumefaciens containing the binary expression vectors, selected based on hygromycin resistance, and regenerated as described in (Zhang et al., 2022). In short, young, transparent embryos were isolated and grown for three weeks on callus induction media (CIM; per L: 4.43g Linsmaier & Skoog basal media (LS; Duchefa #L0230), 30g sucrose, 600µl CuSO4 (1mg/ ml, Sigma/Merck #C3036), 500µl 2,4-D (5mg/ml in 1M KOH, Sigma/Merck #D7299), pH 5.8, plus 2.1g of Phytagel (Sigma/Merck #P8169)). After three weeks of incubation at 28 o C in the dark, crisp, yellow callus pieces were subcultured to fresh CIM plates and incubated for two more weeks at 28 o C in the dark. After two weeks, calli were broken down to 2-5mm small pieces and subcultured for one more week at 28 o C in the dark. For transformation, AGL1 Agrobacteria with the desired construct were dissolved in liquid CIM media (same media as above without the phytagel) with freshly added 2,4-D (2.5µg/ml final conc.), Acetosyringone (200µM final conc., Sigma/Merck #D134406), and Synperonic PE/ F68 (0.1% final conc., Sigma/Merck #81112). The OD600 of the Agrobacteria solution was adjusted to 0.6. Around 100 calli were incubated for at least 10min in the Agrobacteria solution, dried off on sterile filter paper and incubated for three days at room temperature in the dark. After three days, transformed calli were moved to selection media (CIM + Hygromycin (40µg/ ml final conc., Roche #10843555001) + Timentin (200µg/ ml final conc., Ticarcillin 2NA & Clavulanate Potassium from Duchefa #T0190)) and incubated for one week at 28ºC in the dark. After one week, calli were moved to fresh selection plates and incubated for two more weeks at 28ºC in the dark. Next, calli were moved to callus regeneration media (CRM; per L: 4.43g of LS, 30g maltose (Sigma/Merck #M5885), 600µl CuSO4 (1mg/ml), pH 5.8, plus 2.1g of Phytagel). After autoclaving, cool down and add Timentin (200µg/ml final conc.), Hygromycin (40µg/ml final conc.), and sterile Kinetin solution (0.2µg/ml final conc., Sig-ma/Merck #K3253). Calli were incubated at 28 o C and a 16h light:8h dark cycle (70-80 μmol PAR m −2 s -1 ). After 1-6 weeks in the light, shoots will form. Move shoots that are longer than 1cm and ideally have two or more leaves, to rooting cups (Duchefa #S1686) containing rooting media (per L: 4.3g Murashige & Skoog including vitamins (Duchefa #M0222), 30g sucrose, adjust pH to 5.8, add 2.1g Phytagel). After autoclaving, cool down and add Timentin (200µg/ml final concentration). Once roots have formed, plantlets can be moved to soil (consisting of 4 parts ED CL73 (Einheitserde) and 1 part Vermiculite) and grown in a greenhouse with 18h light:6h dark cycle (250-350 μmol PAR m −2 s -1 ). Ideally, the transgenic plantlets moved to soil are initially kept at lower temperatures (day temperature = 22 o C, night temperature = 18-20 o C) for 2-4 weeks until they have rooted sufficiently before being moved to the warmer greenhouse (day temperature = 28 o C, night temperature = 22 o C).
We refer to Brachypodium regenerants as T0 and to the first segregating population as T1. We analyzed 5-10 independent lines in the T0 generation (depending on how many independent lines were recovered upon regeneration). We confirmed the observed expression pattern and performed the phenotyping studies using one to three independent and fertile T0 transgenics that produced seeds (3-4 T1 individuals per line).

Gas exchange phenotyping
Infra-red gas analyser-based leaf-level gas exchange measurements were performed as described in . All measurements were performed on the youngest fully expanded mature leaf of B. distachyon plants 3 weeks after sowing (17-24 days after sowing) using a LI-6800 (LI-COR Biosciences Inc, Lincoln, NE, USA). Ambient light intensity was monitored during the measurements using an external LI-190R PAR Sensor attached to LI-6800. Greenhouse temperature and relative humidity were monitored during the experiments using an Onset HOBO U12-O12 4-channel data logger that was placed next to the plants used for analysis.
Light response kinetics: LI-6800 chamber conditions were as follows: flow rate, 500 µmol s -1 ; fan speed, 10000 rpm; leaf temperature, 28 o C; relative humidity (RH), 40 %; [CO 2 ], 400 µmol mol -1 ; photosynthetic active radiation (PAR), 1000 -100 -1000 -0 µmol PAR m -2 s -1 (20 min per light step). Gas exchange measurements were automatically logged every minute. The leaf section measured inside the LI-6800 chamber was collected, fixed and cleared to measure leaf area and accurately determine A and g sw and stomatal anatomical parameters like stomatal length and density. iWUE was calculated as the ratio of A to g sw . Stomatal opening and closure speed were evaluated by rate constants (k, min -1 ) determined from exponential equations fitted for each of the three light transitions (1000-100, 100-1000 and 1000-0 PAR), as described in .
Anatomical g s max calculations: Anatomical g s max calculations were performed on the 6 individuals for which gas exchange and stomatal anatomy was assessed (Fig. 1D-G). Based on the formula optimized for B. distachyon in  this requires the measurement of three anatomical parameters: stomatal length, stomatal density and GC width at the apex (average of 30 stomata per individual) from images obtained using a Leica DM5000B microscope.
Steady-state stomatal conductance: LI-6800 chamber conditions were as follows: flow rate, 500 µmol s -1 ; fan speed, 10000 rpm; leaf temperature, 28 o C; relative humidity (RH), 40 %; [CO 2 ], 400 µmol mol -1 ; photosynthetic active radiation (PAR), 1000 PAR m -2 s -1 (20 min). Gas exchange measurements were automatically logged every minute. The leaf section measured inside the Li-6800 chamber was collected to measure leaf area to accurately determine A and g sw and, then fixed and cleared to determine stomatal anatomical parameters like stomatal length and density. All measurements were performed in a semi-randomized manner between 11:30 and 17:30 h to assure measurements for each genotype covered identical periods of time of the day and to avoid the influence of the diurnal variation of g sw observed in .
Ambient-adapted stomatal conductance: Steady-state ambient-adapted stomatal conductance was assessed using a SC-1 porometer (Meter, Pullman, Washington, USA). SC-1 was calibrated using the calibration plate and the moist circular filter paper provided with the SC-1. Each measurement was performed in automode (30 s). The relative humidity of the SC-1 porometer was returned to < 10 % after each measurement by shaking the sensor head for 30-90 s. Three to four fully expanded leaves per individual were measured twice. Three B. distachyon (Bd21-3) and three bdpox-1 individuals were assessed three weeks after sowing. All measurements were performed in a randomized manner between 8:00 and 9:30 h, to avoid the influence of the diurnal variation of g sw observed in .
Important note: Some of the wild-type gas exchange measurements were previously published in , where 120 wild-type measurements over the course of 2 years were correlated to variable growth conditions. The WT gas exchange data that was published in  is indicated accordingly in the Supplementary Dataset 2.

Microscopy and Phenotypic Analysis
Most of the morphometric and the cell wall measurements were performed on the actual leaf segments used for gas exchange measurement to thoroughly link cellular form and composition to stomatal gas exchange.
Leaf epidermis morphometry: For DIC imaging, the youngest fully expanded mature leaves (3 weeks after sowing) were collected after LI-6800 measurements and placed into 7:1 ethanol:acetic acid and incubated overnight to fix the leaf tissue and remove chlorophyll. To prepare samples for imaging, the tissue was rinsed twice in water, mounted on slides in Hoyer's solution (Sharma, 2017) and the abaxial side was examined using a Leica DM5000B microscope (Leica Microsystems, Wetzlar, Germany). Typically, 4-6 (40x objective) and 3-5 (20x objective) abaxial fields of view per leaf of each individual plant were imaged to determine stomatal length, stomatal density, stomatal width at the apices, pavement cell length, prickle hair cell (PHC) base length, base area and/or PHC outgrowth using the straight line tool xyz and/or the polygon selection tool in Fiji (Schindelin et al., 2012). In the case of complementation experiments represented in Figure 2, the slides were prepared and randomized by an independent researcher before measurements on Fiji, to avoid potential biased phenotyping. The confocal morphometrical analysis of stomata in Bd21-3 and bdpox-1 mutants, were performed as described in . Leaves were incubated overnight in buffer solution (50mM KCl, 10mM MES-KOH) with 4mM Fusicoccin (Santa Cruz Biotechnology, Inc., Dallas, TX, USA;Cat. no. 20108-30-9) in the light to force stomatal opening. Leaves were stained in propidium iodide (Sigma-Aldrich, St. Louis, Missouri, USA, Cat. no. P3566; PI, 1:100 of a 1 mg/ml stock) for 5 min and Z-stacks were taken using confocal microscopy. Image analysis was done using Fiji to measure stomatal pore length and stomatal pore area (hand-traced).
Stomatal associated-epidermal defects: Stomata and stomata surrounded by defective cell divisions were counted in 5 abaxial fields of view (40 x objective) per leaf of each individual plant. Finally the percentage of stomatal-associated defects was calculated as the total number of stomata surrounded by defective cell divisions using the following formula: sum of 5 fields of view)/total number of stomata (sum of 5 fields of view) x 100.
Reporter lines: For confocal imaging, emerging 2 nd (6-7 days post germination (dpg)) or 3 rd (11-12 dpg) leaves from plants grown on plates were carefully pulled from the sheath of the older leaf to isolate and reveal the developmental leaf zone. Samples were stained in propidium iodide (PI, 1:100 of a 1 mg/ml stock) for 5 min to stain cell walls and/or mounted directly in water for imaging on a Leica SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany). Image anal-ysis was done using Fiji.
Total phenolics autofluorescence: Small leaf fragments previously fixed and cleared in 7:1 ethanol:acetic acid were washed in 70 % ethanol and then transferred to distilled water with 0.02 % (v/v) Tween for rehydration for 3 hours and mounted in distilled water for imaging. Samples were imaged on a Leica SP8 confocal microscope. Excitation and detection settings were as follows: Ex. 405 nm and Em. 490-550 nm. Laser power set to 10 %. For the analysis of PHCs, stacks of 0.33 μm steps were obtained and plot profile analysis was performed on sum slices Z-projections in Fiji. For the analysis of PHCs plot profiles, a straight line was drawn from the base of the PHCs to the tip and plot profiles of gray values were obtained. For the analysis of PHC average autofluorescence, PHCs were handtraced on the sum slices Z-projections and CTCF (corrected total cell fluorescence) calculated as Integrated Density -(Area of selected cell x Mean fluorescence of background readings). Mean fluorescence of 9 background readings per image were obtained to calculate CTCF (using the traced area at regions with only background signal) and to correct plot profiles. For the analysis of mature stomata autofluorescence, GCs were handtraced and autofluorescence was calculated as CTCF (corrected total cell fluorescence) = Integrated Density -(Area of selected cell x Mean fluorescence of background readings) from single images. Mean fluorescence of 3-4 background readings per image were obtained to calculate CTCF (using the traced area at regions with only background signal).
For the analysis of phenolic compounds during stomatal elongation (step 6i-6iii), the emerging 2 nd (6-7 dpg) or 3 rd (11-12 dpg) leaves were carefully pulled from the sheath of the older leaf to isolate and reveal the developmental leaf zone were mounted in water for imaging on a Leica SP8 confocal microscope. Autofluorescence intensity was measured on handtraced GCs from sum slices Z-projections (75 stacks) using Fiji and calculated as CTCF (corrected total cell fluorescence) = Integrated Density -(Area of selected cell x Mean fluorescence of background readings). Mean fluorescence of 3-4 background readings surrounding each stoma were obtained to calculate CTCF for each GC pair.
Basic fuchsin staining: Small leaf fragments previously fixed and cleared in 7:1 ethanol:acetic acid were transferred to distilled water with 0.02 % (v/v) Tween for rehydration for 3 hours. Samples were incubated in 30 µl of 0.01 % Basic Fuchsin (Sigma-Aldrich, St. Louis, Missouri, USA, Cat. no. 857343) for 5 min and washed twice with 30 µl of 50 % glycerol (v/v) for 5 min (2.5 min per wash step), and mounted in 50 % glycerol. Samples were imaged under Ex. 561 nm and Em. 573-603 nm. Stacks of 0.33 μm steps were obtained. For the analysis of PHCs, sum slices Z-projections were performed. A straight line was drawn from the base of the PHCs until the tip and plot profiles of gray values were obtained. Mean gray value of 9 back-ground readings was obtained to correct each measurement. For the analysis of stomata, GCs were handtraced and fluorescence was calculated as CTCF (corrected total cell fluorescence) = Integrated Density -(Area of selected cell x Mean fluorescence of background readings) from single images. Mean fluorescence of 3-6 background readings was obtained for each image to calculate CTCF.
Safranin-o staining: Small leaf fragments previously fixed and cleared in 7:1 ethanol acetic acid were transferred to distilled water for 2 hours. Samples were incubated in 50 µl of 0.2 % Safranin-O (Sigma-Aldrich, St. Louis, Missouri, USA, Cat. no. 84120; v/v in 50 % EtOH) for 10 min, washed with 50 % EtOH for 10 min and hydrated in distilled water for 15 min. Samples were mounted in distilled water and imaged under Ex. 488 nm and Em. 530-560 nm (Channel 1 (C1)) and Ex. 561 nm and Em. 570-600 nm (Channel 2 (C2)). Stacks of 0.33 μm steps were obtained. A straight line was drawn from the base of the PHCs until the tip and plot profiles of gray values for C1 and C2 (same defined ROI) were obtained from the sum slices Z-projections. The ratio between the plot profiles from C2 and C1 was obtained. Additionally, images were analyzed using the Fiji macro developed by Baldacci-Cresp et al. 2020 for calculating a ratio from a generated ratiometric image (C2/C1) (Baldacci-Cresp et al., 2020).